Control.com - Nerds in Control

Is it possible to draw a curve of the GT compressor's fouling with the dat you can have with dump of the memory (Mark V <I>) ?
I have to work on it , but I don't know really where i've to start and the data that are important .

Can someon help me ?
Best Regards

There are several software packages designed to monitor gas turbine- and axial compressor efficiency. They can be quite complex and expensive, but, if configured and used properly they can pay for themselves quite quite quickly. It is recommended to dedicate one person to learn and monitor the software, 'cause generally they ain't for the faint of heart (in other, they take a lotta learnin'--and patience--to realize their full potential).

The big challenge with trying to develop some method to determine when it's necessary to perform a compressor water wash is that it's necessary to compare apples and apples--in other words, the data either needs to be taken at a consistent operating condition (identical compressor inlet temperature with the unit operating at Base Load), or it needs to be corrected for the operating condition at which it was taken back to some consistent point.

If one takes a data point every day at 1200 hrs, the problem is that the ambient inlet temperature isn't the same every day at 1200 hrs, neither is the load (unless the unit is run day in and day out at Base Load during mid-day). The unit must be at steady-state operating conditions (wheelspace temperatures must not be chaning more than 5 deg F in 15 minutes), so the data point cannot be taken as soon as the unit reaches Base Load. If the unit is a load-following unit, meaning that its load is varied during the day, the problem becomes even more difficult.

That's one of the things one pays (handsomely) for when buying performance monitoring software--the ability to correct data back to a particular operating condition so that one can determine the effective-, or relative-, efficiency and then compare that number to the previous day's, week's, month's. Now, these packages use some very sophisticated software to approximate efficiency, some better than others.

The Mark V <I> doesn't have a great method for capturing data, that really is it's biggest downfall. So you'll need to manually capture the data points, input them into a spreadsheet, and then perform your calculations. It is easy to create a User Defined Data Display ("Demand Display") with the data points you want to monitor/capture, then save it to a file or print it when you want to capture the data.

Presuming your unit is Base-loaded each and every day (the best possible condition for this example), you would do best to take several data points each day, at several compressor inlet temperatures, say 40 deg F, 50 deg F, 60 deg F, 70 deg F, 80 deg F, 90 deg F, 100 deg F. Now these temperatures won't all occur every day, but enough of them will to build a database with, and you'll have several points every day.

The data you want to monitor is fuel flow-rate, compressor discharge pressure, compressor inlet temperature, compressor discharge temperature, exhaust temperature, load (power output), ambient temperature, ambient humidity, ambient pressure (barometric pressure).

As you plot these data for each temperature over a period of days and weeks, you will begin to see trends, and you can begin to determine which data points or combination of data points point to declining compressor efficiency--or, a fouling compressor which is the most easily corrected form of degradation (corrected by compressor washing).

The next big challenge is to determine when to shut down to perform an off-line compressor water wash. The economics of your site will determine that point.

The next big challenge is to devise a timely procedure for performing an off-line compressor water wash. The quicker that one can be performed, the sooner the unit can be returned to service--meaning the least lost productivity.

When you perform a compressor water wash, you will (should) see a big change in the next data points, indicating increased efficiency.

But, getting the Mark V <I> to gather data "automatically" is just about impossible, and while it can be electronically collected and saved to a file, it requires a lot of manual file copying and data manipulation to get the data into a format that can be analyzed. It's almost as formidable as trying to get the operators to manually print a particular screen at periodic intervals--nearly impossible.

So, let us know how you fare with your challenge. A LOT of people would be very interested to see what you come up with!

markvguy

Further, since you have an F/FA-class unit, you have compressor pressure ratio (which already takes into account atmospheric (barometric) pressure) and you may also have inlet pressure drop and static pressure drop, and inlet air flow and possibly even exhaust flow--depending on the instrumentation packages purchased and provided with the unit. These signals will help in you quest.

markvguy

Hi, many thanks Markvguy,

As I've said, we're doing a Mark V memory dump 3 times per day (about 7000 data points at a time...), we're taking this data by batch once a week.

Then we put them in a PC; I've made some Xcel Macros and they're sorting all the data (constants in a file, logic in another and last the analogic values)-> we can call that a "database".

Then you can pick what you want in these files.
I'm gonna work with you precious indications and doing a file sorting the compressor data that's interesting me for the compressor fouling.

Say you what I've done in a few days.

Best regards,
laurent_f

Laurent_f,

My answer is a bit late as I only saw the questions today.

It's easy to do in Excel. Download the data into Excel and write a sub to eliminate all VIGV settings less than say 98 to 100% so that you can eliminate readings less than around 100% load. Tabulate these readings in date and time order columns and in adjacent columns log the corresponding ambient temp, RH pressure. Develop a lookup table for the engine correction curves and correct the output back to the site design standard or ISO or whatever you choose as a reference. Then plot the corrected output data with time and you have an output degradation curve due mainly to fouling (it includes filter DP as well).

If you want you can take the fuel flow readings and calculate the heat rate on the same basis (don't forget to correct it to your standard conditions) and plot it against elapsed time then you also have a HR degradation curve with time.

Alternatively you can pay me to do it as I do this stuff for a living.

Hi, Thanks to all.

I've done a Excel macro to order the data, I'm trying to organize the corrections curves. -> Lot of other works... Take times. It begins to have a look of something. ;-)

Laurent

In reply to laurent_f: If the files are in ASCII format, you might look at using "AWK" to process them. AWK is a simple programming language meant for exactly this type of application. Most of the work of getting the data into and out of the program, looping through it line by line, and splitting the data into fields are done in the background by the interpreter. You just have to worry about writing your filter logic (usually just a series of "if" and "print" statements).

I had some routine text file processing operations that I originally did in MS-Excell which I later converted to using AWK. The MS-Excell solution sounded "easy" at first sight, but it got complicated very fast while still being very limited (it was also prone to errors, due to the complexity).

I converted it to AWK, and was able to do a lot more with the data and do it much more easily and much, much faster. I was also able to set it up as a batch operation using batch files so that I could process several dozen files automatically in seconds. The resulting output was still imported into a spreadsheet for display purposes and to allow ad-hoc manipulation of the data.

AWK comes from several sources. By far the most common one these days is GAWK from GNU. It's free of course, and there are versions for most operating systems.

If your filter logic is really simple (just sorting out lines of text), then something like GREP might do the job as well.

What is compressor fouling??

Regards
Grishma

The axial compressor of a combustion turbine gets dirty, despite the best air filters; this is called "fouling." The dirtier it gets, air flow through the compressor. Less air flow, less power output.

markvguy

Air intake Filters are not all the same, and "best" doesn't really apply to the OEM supply. Therefore, if you are still utilising Donaldson pulse filters in non-sandstorm environments, you definetely cannot refer to them as good filters.

gdearcangelis@tteur.com